Excited state dynamics of carbonyl carotenoids investigated by ultrafast vibrational and electronic spectroscopies.

(1)

(2)

Introduction

In the last decades private industries as well as public funds have supported many laboratory research activities on solar energy all over the world, in order to provide an alternative source of energy to carbon fossils and satisfy the increasing energy demand. The overall amount of energy the Sun leaves on Earth is well above human consumption, but its conver-sion in electricity is limited to the visible spectral range and at has low efficiency. Likely, solar power could never provide all the electrical supply each country needs, but in combi-nation with other green and renewable resources like wind, geothermal gases, biomasses and hydroelectric power a re-markable decrease of carbon fossil exploitation is possible.

Solar energy is nowadays transformed in several ways: from photovoltaic (PV) devices, in which the radiative energy is directly transformed in electricity, to thermal systems, where sun light collectors increase the temperature of a mineral oil up to gasification, thus activating turbine engines for

(7)

elec-Introduction

tricity production. Nowadays commercial PV panels, made of mono-crystalline Si wafers, have an efficiency (calculated as produced electrical power divided by the absorbed sun energy intensity per panel unit) close to 15% and an aver-age lifetime of 30 years.1 _{Despite the large-scale production,}

their costs are still not competitive with carbon fossil fuels, thus further improvements on their efficiency and long-time stability are needed.

Several hypothesis are under study.2_{On the inorganic side,}

improvements are expected from the substitution of the Si wafers with thin films of less expensive CdTe and CuInSe2

and from the modification of the solar cell morphology, from single wafer to multi n-p junctions (presently the best labora-tory efficiency is around 30%). However, the most promising alternatives in transforming sun energy mix the organic and the inorganic fields. Hybrid technologies involve new knowl-edge in nano-science, conductive polymers and bio-molecular science. A deep knowledge on the electronic and dynamical properties of natural pigments may help in mimicking the natural photosynthesis processes of light-harvesting, energy transfer, charge separation and energy storage3,4_{. If natural}

or artificial dyes, with optimal characteristics in all these pro-cesses, were found, they could be used as sensitizer in poly-meric or solid semi-conductive matrices, as already imple-mented in Gr¨atzel solar cells to functionalize nano-particles of TiO2.5 The first dye-sensitized solar cells (DSSC) have

(8)

Introduction been realized in 1991 and their production has been rapi-dly industrialized because of their lightweight and flexibility in terms of materials they are made of. Dye-sensitized ma-terials could find in a close future wide application: from small rechargeable electronic devices to eco-friendly houses and self-powered design elements.

In this Ph.D. thesis, the attention has been focussed on the electronic properties and the relaxation dynamics of a selected class of natural pigments: the carbonyl carotenoids. Carotenoids are present in all antenna protein complexes and act mainly as accessory light harvesting pigments of chlorophylls, since they absorb the blue-green visible region of the solar spectrum, or as photo-activity regulator and photo-protector against free radicals and harmful singlet oxy-gen. In many marine algae proteins, however, a larger num-ber of carbonyl carotenoids than chlorophylls has been de-tected, like in the Peridinin-Chlorophyll a-Protein (PCP) complex of the dinoflagellate Amphidinium Carterae6_,

sug-gesting a key-role of this pigment in light-harvesting and a high efficient energy transfer to the photosynthetic reaction center. These simple organisms live in low-illumination con-ditions, and their photosynthetic apparatus has evolved to absorb in the blue-green spectral region, being the red-most part of the solar spectrum strongly absorbed by water. An energy transfer efficiency approaching the 90% has been

(9)

ob-Introduction

served in PCP,7–9 _{so that there is an increasing interest in}

exploiting carotenoids as dye-sensitizers or as donor groups in artificial push-pull compounds. However, the photo-dynamics of carbonyl carotenoids is not yet fully understood. Contrar-ily to non-polar carotenoids (such as the all trans β-carotene, see figure), the excited state lifetime of carbonyl carotenoids is very sensitive to the polarity of the environment. Further-more, many of the naturally occurring carbonyl carotenoids present more than a polar functional group, thus complicat-ing the symmetry of their chemical structure and the elec-tronic properties of their excited states.

In case of natural antennas, significant insights about the role of the protein environment in determining the energy transfer properties of the system was gained from high tem-poral resolution pump-probe measurements. It is known from previous studies,10_{that specific pigment-protein interactions}

as well as the dielectric heterogeneity of the protein domains are important to tune the pigment site energies and influ-ence the energy transfer. The high level of organization in terms of displacement and orientation of the chromophores within proteins and of the antennas around reaction centers, certainly, has an influence on the observed high efficiency of energy transfer.

In this thesis, as a preliminary step towards the analy-sis of the photo-dynamics of Peridinin, we considered a

(10)

sim-Introduction pler molecule, the trans-β-apo-8’-carotenal. This carotenoid presents a single carbonyl group, what allows us to investi-gate the effects of the polyene chain symmetry breakdown on the excited state relaxation dynamics, and to clarify the influ-ence of the external medium polarity. Then we analyzed the photo-dynamics of Peridinin in several solvents, using time resolved spectroscopies mostly based on the use of infrared pulses. H O O O O O O _OH OH H O O O O O O _O H H O O O O O O O H H O O O O O O _OH

1) all trans beta-carotene

2) all trans beta-apo-8’-carotenal

(11)

Introduction

The thesis is organized as follows: a brief description of the photosynthetic apparatus and of the pigment properties is reported in chapter 1; the second chapter outlines some the-oretical aspects regarding the spectroscopic techniques and describes the instrumental set-ups. The experimental results on the analyzed carbonyl carotenoids, namely the all trans-β-apo-8’-carotenal and Peridinin, are presented and discussed in chapters 3 and 4, respectively. Finally, a model that ex-plains the experimental evidences, including polarity and po-larizability solvation effects, is proposed in chapter 5.

(12)

1 The Photosynthesis

6 CO2+ 12 H2O

photosynthesis

cellrespiration C6H12O6+ 6 O2+ 6 H2O

Scheme 1

Photosynthesis is the basic process allowing plants and mi-croorganisms (e.g. algae and bacteria) to convert light into chemical energy. The photosynthetic organisms are able to store the harvested light in the chemical bonds of ATP, which drives the synthesis of glucose, essential for their sustainment. In this way, they fix carbon atoms of CO2 molecules of the

air and deliver O2 as a side product. The overall process

is a reduction-oxidation reaction, in which carbon atoms of CO2 molecules are reduced and oxygen atoms of water are

oxidized. The opposite eso-ergodic reaction is the cellular respiration, in which the biochemical energy of nutrients (the glucose) is converted in ATP, NADPH or equivalent cell

(13)

en-1. The Photosynthesis

ergy carriers. In both processes, the overall chemical reaction shown in scheme 1 hides several steps.

The photosynthesis is usually divided in two stages: in the first light-dependent reactions stage several pigments harvest sunlight and funnel the energy to the reaction center, where the first photo-induced charge separation starts the electron transportation chain. The latter process ends up in the re-duction of a NADP+ _{molecule and in the activation of the}

ATP-synthase by the transmembrane electro-chemical poten-tial. In the second dark reactions stage, these molecules pro-vide the energy to fix three CO2 molecules in the

glycer-aldehyde 3-phosphate (G3P), the metabolic intermediate of a glucose molecule. This second stage is also known as Calvin Cycle or C3 cycle in plants and green algae. In the following, we will focus on the first stage.

(14)

1. The Photosynthesis Stroma' Lumen' The$Thylakoid$membrane:$ Stroma Lumen Stroma' Lumen'

The$Thylakoid$membrane:$_ThylakoidThe

membrane:

Figure 1.1: Schematic view of a chloroplast and the thylakoid membrane.

Photosynthetic reactions occur inside chloroplasts (figure 1.1), the engine-subunits of vegetal cells, like mitochondria in animal cells. This organelle has two external membranes which contain a semi-gel fluid called stroma, where thylakoid membranes organized in grana and lamellae float. While the Calvin cycle takes place in the stroma, most part of the bright reactions happen in embedded proteins of the thylakoid mem-brane. These large protein complexes are, in the order, the

(15)

1. The Photosynthesis

photosystem II (PSII), a cytochrome complex responsible of the transmembrane electrochemical potential, the photosys-tem I (PSI) and the ATP-synthase. The cytochrome is di-rectly involved in the electron transportation chain from PSII to PSI, together with the plastoquinone pool and the plas-tocyanin protein. This redox reactions cascade activates a protonic pump which transfers H+ _{from the stroma to the}

inner thylakoid space, the lumen. As a consequence of the electrochemical gradient between the two faces of the mem-brane, H+_{ions pass back to the stroma through the channel}

of the ATP-synthase, thus activating the phosphorylation of an ADP molecule. In this way, part of the initial photon energy, which seems to be lost along the electron chain, is recovered.

Figure 1.1 reports the characteristic scheme of the thy-lakoid membrane of the oxygenic photosynthesis of green plants, however, in many algae and bacteria, photosynthetic reactions take place even in the absence of water, thus with-out producing oxygen. Photosystems I and II are very sim-ilar protein complexes, composed of many antenna systems and coenzymes surrounding the central reaction center (RC) protein. Antenna complexes are rich of pigments, such as chlorophylls and carotenoids, which harvest light and trans-fer energy to the reaction center. Here, the so-called special pair P680 or P700 (respectively in the PSII and PSI of green plants), consisting of a dimer of Chl-a molecules, is excited

(16)

1. The Photosynthesis through energy transfer from nearby antennae or by direct absorption of photons at 680 and 700 nm. This energy acti-vates a charge separation process, which finally results in the oxidation of the special pair and the reduction of a quinone molecule. The D1 and D2 domains of the RC protein are highly symmetric, with two pathways for electron transfer almost equivalent, however only the D1 domain results to be active.11 _{Once excited, the special pair P680 transfers the}

electron to a Pheophytin (Ph) molecule, which has the same porphyrin structure of a Chlorophyll molecule with the cen-tral Magnesium ion replaced by two protons; then the charge passes to a primary tightly protein-bound quinone (QA) and

finally to a second exchangeable one (QB). At each round,

QB gains a negative charge and binds an H+ ion taken from

the stroma, increasing the electro-chemical transmembrane gradient. After two rounds, it leaves PSII as QH2 and

dif-fuses through the membrane to the cytochrome complex. The re-oxidation of QH2 provides the energy for the

transmem-brane protonic transfer and the charge to oxidize a copper atom of the plastocyanin protein, which transports the elec-tron to PSI. Here, the special pair P700 starts a sequence of electron transfer processes similar to those occurring in PSII. The charge finally proceeds to the Ferredoxin protein, attached to PSI on the stroma side, where the reduction of a NADP+ _{molecule takes place. Up to now, it has not been}

(17)

first photo-induced separation charge in the P680+_Ph−

com-plex. Even though thermodynamically favorited, the charge recombination does not take place because Ph−_{transfers the}

electron to the close quinone QA faster than recombination.

The remaining P680+ _{is a strong oxidant, which extracts}

an electron from a protein attached on the lumen side called oxygen-evolving complex. The active site of this protein con-tains 4 atoms of Mn at different oxidation state and coordi-nates two water molecules. The electron extraction forced by the special-pair catalyzes the water splitting: every four electrons an oxygen molecule is produced and four H+ _are

released on the lumen side.

In photosynthetic bacteria the thylakoid membrane struc-ture is homologous to that of green plants, however the num-ber of cofactors and coenzymes in each photosystem is smaller. Many antenna and reaction center photo-systems have been determined at atomic resolution.12 _{(and ref.}

139-140,154-156,159 therein). In purple and green bacteria D1 and D2 do-mains are historically named as M and L branches, with L as active branch. Bacterial special pairs are dimers of bacterio-chlorophylls (BChl), which absorb around 800-900 nm, in a red-shifted region with respect to P680 and P700, likely due to stronger pigment-protein interactions in the bacterial RC. In the first photo-induced separation charge, a BChl mediates the charge transfer from the special pair to the bacterio-pheophytin (BPh).13 _{The striking difference from}

(18)

1. The Photosynthesis green plants is the way the electron is replaced in the special pair after the photo-induced separation. No water splitting coenzyme are attached to the reaction center, however a cy-tochrome type c provides the needed electron at each cycle, avoiding the charge recombination.

1.1 The Light-Harvesting Proteins

As seen in the previous paragraph, the PSII and PSI super-complexes contain many proteins with different functionali-ties: from cytochromes involved in the electron chain trans-port up to the core reaction center surrounded by many light-harvesting (LH) antennae, which contain several chlorophylls and carotenoid chromophores. It has been observed through atomic force microscopy14 _{that the number of the antenna}

systems per reaction center is strictly correlated with the available light and increases in low-light conditions, together with the overall cross-section of photon absorption. Indeed, in order to fully reduce the quinone QB two electrons are

needed. Therefore, the photo-excitation rate of the RC must exceed a certain threshold to completely avoid charge recom-bination.

In order to optimize the light-harvesting, in highly evolved organisms such as green plants, antenna proteins can be

(19)

dis-1. The Photosynthesis

tinguished in peripheral (LHCII and LHCI, respectively in the PSII and PSI), semi-peripheral (CP29, CP26 and CP24) and core antenna (CP43 and CP47) (figure 1.2), with differ-ent value of energy transfer yield toward the RC.10,12

Bac-terial photosystems are less complex, thus light harvesting proteins are simply distinguished in LH2 and LH1 proteins for the two photosystems. The antennas are either embedded across the phospholipid thylakoid membrane, or dissolved in water and chemically anchored on the lumen or on the stroma side of the membrane. Moreover these systems present a great variety in their 3D structures as well as in the inner disposition of their pigments (figure 1.3).

LHCII CP24 CP26 CP29 RC CP43 CP47 LHCII

Figure 1.2: Light Harvesting Proteins in the Photosystem II su-percomplex of green plants.

(20)

1. The Photosynthesis LH2 (Rhodomonas acidophila) PCP (Amphidinium Carterae) LHCII (Spinacia olearia)

Figure 1.3: Structural variety of antenna proteins.

LH protein Carotenoids

LHCII of green plants Beta-carotene Lutein Violaxanthin

Neoxanthin Zeaxanthin LH2 Rhodomonas acidophila Rhodopin glucoside PCP Amphidinium Carterae Peridinin

(21)

In light-harvesting complexes, chlorophylls are generally present in larger number than carotenoids, which mainly act as photo-protector against harmful oxygen singlet and chloro-phyll triplet states15–17 _{. One of the most studied}

photo-protective mechanisms is the non-photochemical quenching (NPQ)18–20_{in which the Chl fluorescence is quenched by the}

Xanthophyll Cycle. Here, a sequence of reactions triggered by the high-light conditions, converts violaxanthin in zeaxan-thin, thus dissipating the excess of energy as thermal energy. As reported in table 1.1, in plants different types of caro-tenoids are present within the LHCII protein21_{. They}

exe-cute different functions and are not necessarily optimized for light-harvesting and energy-transfer. In bacteria, the type of carotenoid contained in LH complexes is different for each species. Since only one type of carotenoid is present, it is usually able to efficiently absorb light and transfer energy to BChlorophylls. The determination of the ET yield is here easier than in plant antennae: selective excitation of carote-noids is accessible since the visible absorption band around 400-550 nm results to be narrower and generally does not overlap with the BChl absorption below 400 nm (the so called Soret band). As reported in greater details in the next para-graph, the Car-Chl ET proceeds from both the first and the second carotenoid excited states, however, while in LHC pro-tein family the main transfer pathway (∼ 80%) is from the second excited state, in marine algae it originates from the

(22)

1. The Photosynthesis first excited state. Different type of ET mechanisms have been hypothesized accordingly to the different analyzed an-tenna systems.

1.1.1 The Peridinin-Chlorophyll a Protein

Figure 1.4: Trimeric structure of the water soluble PCP protein (PDB-ID: 1PPR).

(23)

Here we focus the attention on the Peridinin-Chlorophyll a Protein (PCP) reported in figure 1.4. The uniqueness of the PCP protein is in its chromophoric composition. The PCP protein is a water soluble protein extracted from the dinoflag-ellate Amphidinium Carterae, a unicellular organism that lives in under-illuminated conditions in tropic coastal waters, commonly found as phyto-plankton up to 20-30 m in depths. This protein contains only one type of carotenoid, the Peri-dinin, and in a larger number than Chlorophyll molecules. Therefore, Peridinin acts as primary light absorbers and trans-fers the energy to chlorophylls with a yield which has been found to reach 90% .7–9 _{The X-ray structure of this protein}

was determined6 _{at high resolution (2Å) in 1996; it reveals}

a trimeric form in which each monomeric unit contains two identical pseudo-twofold symmetric domains: in each domain 1 Chl-a and 4 Peridinin molecules are in Van der Waals con-tact through Peridinin conjugated chains and Chl-a tetrapyr-role ring (figure 1.5). Furthermore, Peridinins are two-by-two H-bonded across the two domains. It is worth pointing out that the quaternary structure of the all protein is based on a central water molecule, which coordinates three Peridinin molecules from each monomer through the OH groups placed on the terminal epoxy-ring.

(24)

Figure 1.5: Symmetric two-fold disposition of 2 Chlorophyll-a and 8 Peridinin molecules in a monomer of PCP.

1.1.2 Energy-Transfer mechanisms in

Light-Harvesting Proteins

In natural antennas pigments are generally geometrically arranged in the protein scaffold, but even when they seems to be randomly disposed their orientation and distance are optimized for ET.

Two mechanisms are generally invoked to explain ET: the F¨orster mechanism (or FRET, F¨orster Resonance Energy

(25)

Trans-1. The Photosynthesis

fer), active when the inter-chromophoric distance is between 1 and 10 nm, and the Dexter mechanism, operating when molecules are in close Van der Waals contact (<10Å). Both mechanisms are non-radiative and their efficiency depends on the overlap integral of the donor fluorescence and acceptor absorption spectra. The substantial difference is the strength of the electronic coupling between the energy donor and the acceptor. In the F¨orster mechanism each molecule maintains its own electrons. This mechanism is based on the resonance between a donor and an acceptor molecule, which are dipole-dipole coupled. The maximum ET efficiency is reached when the fluorescence transition dipole moment of the donor and the absorption dipole moment of the acceptor are parallel, thus, generally, a geometrical disposition among pigments is preferred, considering also the 1

r6 dependence on the inter-chromophoric distance. In the Dexter mechanism, instead, molecules are in Van der Waals contact and a physical ex-change of electrons between donor and acceptor molecules, in the excited and in the ground states, is possible.

Both mechanisms correspond to an incoherent hopping of energy, through electronic displacement or electron energy excitation resonance. Only very recently, the presence of quantum-mechanical electronic coherences has been observed thanks to development of new techniques as 2D-Electronic Spectroscopy (2D-ES). Analysis of coherent energy transfer in many photosynthetic systems is actually a cutting-edge

(26)

1. The Photosynthesis topic.22–25

1.2 Electronic properties of

Carotenoids

Carotenoids are naturally occurring pigments character-ized by a long conjugated carbon chain with conjugation length N = 7-13. They can be classified in xanthophylls, con-taining oxygen atoms, and neutral carotenes. Excited state properties of carotenoids largely depend on the chain con-jugation length, on their molecular symmetry, on the pres-ence of terminal β-rings and attached side-groups. The sim-plest energy level scheme used for carotenoids is based on the electronic properties of all-trans-polyenes with conjugation length N≥ 7 (see Appendix)26,27_{. Their C}

2h symmetry leads

to a first singlet S1(2A−g) excited state, whose direct

excita-tion from the S0(1Ag−) ground state is forbidden by selection

rules. The visible band between 400 and 550 nm is therefore ascribed to the second singlet S2(1B+u) excited state, which

rapidly relaxes via conical intersection to S1 within a few

hundred femtosecond (figure 1.6)28_.

Together with chlorophylls, carotenoids cover almost all the visible spectral range of the incident solar radiation on Earth. It is worth noting that carotenoids, absorbing the

(27)

green-blue light, cover the highest efficient part of the sun radiation scattering across deep water: Iscattering ∝ _λ14. Fig-ure 1.6 reports the absorption spectrum of a typical photo-synthetic antenna. Besides carotenoids bands, Chlorophylls absorptions are also identifiable: a doublet Qy peak around

700-850 nm, whose frequency position shifts according to the strong coupling between chlorophyll molecules arranged in different position inside the protein, a Qx band around 600

nm and the Soret peak in the UV spectral region. In many cases, the carotenoid-chlorophyll ET yield is close to unity: as already mentioned, both S2and S1carotene excited states

can be involved in the energy transfer, however there is also computational and experimental evidence that other dark states could take part to the ET, especially in the pigment-protein bound form.23,27,29–31

(28)

1. The Photosynthesis S2 S1 S0 Carotene Chlorophyll Qx Qy S0 ≈200 fs <1ps <300 fs ET 1-200 ps CT ≈3 ps a) b)

Figure 1.6: a) From ref.12: Vis absorption spectrum of the LH2

complex of Rps. Acidophila. Horizontal bars show spectral re-gions of carotenoid (orange) and chlorophyll (purple) absorption

bands. b) Possible ET pathways from carotene to chlorophyll

molecules. The CT arrow and the 3 ps time constant are proper of the charge separation in the special pair-Pheophitin complex.

To analyze the role of dark states in the ET and in the re-laxation dynamics of carotenoids many quantum-mechanical (QM) calculations on carotenoid excited states have been car-ried on over decades, with increasing computational power and technical sophistication.27,32 _{QM calculations pose}

sub-stantial problems due to the very high electron correlation along the polyenic chain. When multiple excited configura-tions (IC) at the Franck Condon geometry are considered, S1(2A−g) and S2(1B+u) states loose their specific covalent and

(29)

con-1. The Photosynthesis

figurations. Considering only singly and doubly excited con-figurations is enough to push the 2A−

g state below the 1B +

u in

carotenoids with N≥7. On the contrary, only if higher orders of interaction are included, the 1B−

u and the 3A−g dark states,

which supposedly lie close in energy to S2 and S1, exceed in

energy S2.33

Excited state wave-functions can be visualized in terms of molecular orbitals through the predominant configuration: in this sense the S0 → S2 transition can be described as

a single electron promotion from the HOMO to the LUMO state, while the S0→ S1 transition as a double excitation of

two electrons to the same state (note that the S1absorption

is indeed two-photon allowed).34,35

Non-linear spectroscopies, like two-photon absorption or the pump-dump-probe technique, allow for the direct energy determination of S1state. First attempts to energetically

lo-calize S1 were carried out by measuring Raman excitation

profiles.36 _{The visible excitation wavelength was scanned,}

looking for the amplified Resonant Raman signal of the C=C symmetric stretching. Steady-state fluorescence quantum yield and time-correlated single photon counting experiments were also performed.37,38 _{The fluorescence quantum yields of}

ca-rotenoids in solution are on the order of 10−5_{, because of the}

forbidden nature of the S1-S0 transition, and decrease with

(30)

1. The Photosynthesis a clear switch to the S2 fluorescence was observed37,

there-fore a violation of the Kasha’s rule. Both S1and S2 energies

decrease with increasing conjugation length, however the S1

decrease is slightly steeper, so that the S2-S1energy gap

in-creases with N and the S2 fluorescence overcomes that from

S1. Nevertheless, non-radiative S2-S1and S1-S0internal

con-versions, which follow the energy gap law38_{, represent the}

(31)

1.2.1 Carbonyl Carotenoids

While non-radiative relaxation dynamics of non-polar caro-tenoids are insensitive to solvent polarity, in carbonyl carote-noids the S1lifetime strongly decreases in polar solvents,

go-ing from hundreds of picoseconds to a few picoseconds.39–43

The solvent dependence of the fluorescence lifetime of Peri-dinin was initially attributed to the presence of the lactone ring. However, the dependence of the S1 lifetime on the

sol-vent polarity was observed also in apocarotenals and dicyano-apocarotenes and attributed to the presence of an electron withdrawing substituent on the conjugated chain in an asym-metric position.44,45

Transient absorption measurement on a series of apoc-arotenals, differing for the conjugation length N, showed that the S1lifetime (τ) dependence on the polarity of the solvent

increases with decreasing N.46,47_{Going from non-polar to}

po-lar solvents, τ decreases from 200 to 8 ps in apo-12’-carotenal (N=6), and from 25 to 8 ps in apo-8’-carotenal (N=8). Fi-nally, no solvent polarity dependence (τ = 4-5 ps) is observed in apo-4’-carotenal (N=10). As for non-polar carotenoids, the internal conversion rate between low-lying excited states and the ground state increases with the chain length.

Dual fluorescence is also observed for carbonyl

caroteno-ids;46,48,49 _{it shows a predominance of the S}

(32)

inde-1. The Photosynthesis pendently of the conjugation length. The fluorescence from S1 increases with respect to that from S2 at increasing

sol-vent polarity,49_{although involved S}

1and S0states should be

totally symmetric and insensitive to the solvent property in an idealized C2h symmetry.

All these observations suggest the breakdown of the 2Ag−

symmetry in S1and the presence of an intramolecular charge

transfer (ICT) character, induced by the electron withdraw-ing substituent. In polar carotenoids the low-lywithdraw-ing state is generally referred as the S1/ICT state, however its exact elec-tronic nature still remains unclear. Two are the hypothe-sis proposed: the existence of a separate dark singlet ICT state, coupled to the 2A−

g state and whose molecular nature

does not rise from the C2h polyenic frame, or the existence

of a low-lying 2A−

g/1B+u strongly mixed state, whose

cova-lent/ionic nature is solvent dependent.33,50–52

Characteristic bands of the ICT state, whose band shape and evolution is sensitive to the solvent polarity, are local-ized in two regions of the transient visible spectra. Around 600-700 nm a broad positive signal is assigned to ICT → Sn excited state absorption; at 900-950 nm a negative bleaching band, which grows in few picoseconds, is assigned to the ICT →S0 stimulated emission.42,53 _{The effects of solvent}

polar-ity, viscospolar-ity, temperature, excitation wavelength54_,

(33)

vis-1. The Photosynthesis

ible spectra of carbonyl carotenoids have been largely studied and several hypothesis on the relaxation dynamics have been reported. Different electronic level scheme have been sug-gested, like, for instance, the one proposing an initial branch-ing from S2populating two S1and ICT separate states.

How-ever proposed models mainly diverge for the involvement of different hot states and for the presence of possible equilib-rium dynamics between S1 and ICT.47,57 The low-lying

po-tential energy surface of carbonyl carotenoids turns to be quite complex: the simultaneous presence of the 2Ag−

ex-cited state absorption band around 500 nm and that of the ICT transient bands in polar solvents demonstrates the exis-tence of at least two local minima out of the Franck Condon region.

Infrared and Raman spectroscopies, being sensitive to molec-ular rearrangements, should provide more insights, however only limited transient Infrared and Raman measurements on carbonyl carotenoids are reported in literature.58–60 _{One of}

the most informative infrared frequency region is that around 6-7 µm, where the C=O (∼1650 cm−1_{) and C=C (∼1550}

cm−1_{) stretching modes fall. The frequencies of the}

vibra-tional modes in excited states can be higher or lower than in the ground state, depending on the vibronic coupling with the specific electronic transition. From time resolved Reso-nant Raman measurements,61 _{later confirmed by}

(34)

1. The Photosynthesis stretching mode was observed to shift from 1525 cm−1 _in

the ground state up to 1750 cm−1 _{in S}

1, both in the

non-polar β-carotene and in the carbonyl β-apo-8’-carotenal, be-cause of the strong vibronic coupling between the symmetric S0(1Ag−) and S1(2Ag−) states. Also in the steady state

Res-onant Raman spectra, the band shape and intensity of the excited state C=C stretching mode of β-apo-8’-carotenal re-sulted very sensitive to the solvent polarity. Due to the low molecular symmetry, this mode is infrared active too, how-ever the literature on the solvent dependence of this band is quite limited.

In the next chapters, we will present new measurements based on the use of infrared femtosecond pulses, and we will discuss extensively the possible localization of fingerprint vi-brational modes in the excited states.

(35)

2 Spectroscopic

techniques and set-ups

2.1 Steady-state UV-Vis and IR

Absorption

All the steady-state Uv-Vis absorption spectra reported in this thesis have been recorded with a Perkin Elmer Lambda 950 spectrophotometer at 1 nm resolution. IR spectra, ac-quired on a Bruker Alpha-T FT-IR spectrophotometer at 0.2 cm−1 _{resolution, have been used as calibration reference of}

transient infrared spectra. The sample integrity was checked by infrared and visible absorption before and after all time-resolved measurements.

(36)

2.2 Transient UV-Vis Absorption Spectroscopy

2.2 Transient UV-Vis Absorption

Transient UV-Vis absorption spectroscopy (TAS) is an op-tical pump-probe technique that makes use of ultrashort laser pulses to investigate reaction paths, energy and electron trans-fers and excited state relaxation dynamics. In this exper-iment a first short and energetic beam promotes molecules from the ground to a specific electronic excited state. At different time delays, a second broad and weaker pulse, i.e. the probe pulse, which extends from 350 to 750 nm, moni-tors the system evolution. The probe arrival measures the bleaching of the ground state, stimulates emission from ex-cited molecules or it can be further absorbed toward higher electronic resonant states. Since transient spectra are usually plotted as absorption difference spectra

∆A=ApumpON-ApumpOF F

positive absorbance changes are associated with excited state absorption dynamics, while negative signals correspond to bleaching and stimulated emission phenomena. All dy-namics are reconstructed by recording transient spectra at different pump-probe time delays.

(37)

2.2.1 Set-up

The signal measured in a pump-probe experiment origi-nates from the 3rd_{order polarization, and depends}

quadrati-cally on the intensity of the pump and linearly on that of the probe. In order to have high temporal resolution and large tunability of the excitation wavelength, which is usually gen-erated by non-linear parametric amplification, a short pulsed laser source with high peak power is needed.

Ti:Sapphire Laser System λ=800 nm; E = 500 µJ; Δt = 100 fs; 1 KHz SHG in a BBO crystal Double Array Delay Line CaF2 Ref. White Continuum Generator Probe Sample Holder TOPAS (OPA) 540 nm 400 nm Computer

Figure 2.1: Transient UV-Vis absorption set-up.

Transient visible spectra reported in this thesis were ac-quiredemploying a femtosecond Ti-Sapphire laser system (BMI

(38)

2.2 Transient UV-Vis Absorption Spectroscopy Alpha 1000), which produces 100 fs pulses, centered at 800 nm (FWHM 20 nm), with 500 µJ energy per pulse, at 1kHz repetition rate (figure 3.5).63,64 _{All measurements were}

car-ried out at two excitation wavelengths, in order to be on the blue and red edge of the static absorption spectrum of carotenoids in the different analyzed solvents. The "blue" ex-citation was set at 400 nm independently on the solvent and was generated by frequency doubling a portion of the laser fundamental in a BBO crystal. The "red" excitation was in-stead finely tuned between 520 and 540 nm, accordingly to the solvent, and was obtained as the sum frequency of the signal output of a commercial optical parametric amplifier (TOPAS, Light Conversion) with a portion of the fundamen-tal output of the laser. The probe white light continuum was generated by focusing a fraction of the 800 nm funda-mental on a CaF2 window. It was split into two equally

intense beams by a 50/50 beam splitter; the beam acting as the probe was spatially and temporally overlapped with the excitation beam in the sample by a parabolic mirror in an almost collinear scheme, while the second beam was delayed to provide a convenient reference signal. Finally, probe and reference beams were spectrally dispersed in a flat-field 25 cm Czerny-Turner spectrometer, and detected by means of a back illuminated CCD camera with spectral response in the region 350-750 nm. A moveable delay line made it possible to vary the time-of-arrival-difference of the pump and probe

(39)

beams up to 2.0 ns. The pump beam polarization was set to magic angle with respect to the probe beam by rotating a λ/2 plate. In order to avoid photo-degradation, the pump energy was adjusted at 150-200 µJ at the sample. This latter was continuously stirred by a micro-magnet held in a 2 mm quartz cuvette.

2.2.2 Data Analysis

bl ea chi ng S₀ S1 S2 SE ESA pum p ESA bleaching, SE 500 550 600 650 700 -0.10 -0.05 0.00 0.05 0.10 Δ O D wavelength (nm)

Figure 2.2: Transient UV-Vis absorption signals: positive ex-cited state absorption (ESA) and negative bleaching and stimu-lated emission (SE) signals.

In the measurement, part of the probe light is scattered by the sample, while the pump pulse naturally induces

(40)

flu-2.2 Transient UV-Vis Absorption Spectroscopy orescence. Photons from both undesired phenomena enter into the spectrometer. In order to correct for these effects, reference spectra are acquired in the absence of the pump (to correct for the probe light scattering) and in the absence of the probe (to correct for the fluorescence), before starting the experiment. Home-written LabVIEW software manages the large amount of data and saves the corrected and aver-aged spectra for each pump-probe delay. The two-way matrix (spectral intensity as a function of wavelengths and pump-probe delays) is then reconstructed in Matlab and serves as the input of the GLOTARAN package (http://glotaran.org)65,66

used for the global analysis of the data. In the global anal-ysis process, spectra at different delays together with kinetic traces at different wavelengths are simultaneously fitted to extrapolate time-independent spectra and wavelength inde-pendent kinetics.67_{As the first step, the single value}

decom-position (SVD) procedure extracts the number of significant components from the diagonalization of the input matrix. These components can be seen as compartments (e.g. ex-cited states) that have to be linked through a convenient kinetic scheme. Applying a preliminary sequential first order kinetic scheme, one obtains the Evolution Associated Differ-ence Spectra (EADS, or simply EAS), which are indicative of each compartment and highlight the featuring evolving bands. Often, however, EADS are not the representative spectra of the species involved in the photo dynamics of the

(41)

analyzed system. Departing from EADS, a more complex ki-netic scheme can be applied in the so-called target analysis, to obtain the Decay Associated Difference Spectra (DADS), which are real spectra associated to compartments. However, if the kinetic scheme of the studied system is not known from other experiments, the application of the target analysis can bring to wrong interpretation of data. In the Glotaran fitting algorithm, several parameters are adjustable according to the technical characteristics of the set-up, like the position and the width of the instrumental function and the group velocity dispersion (GVD) of the "white" probe, which is fitted by an adjustable polynomial function. The experimental zero time (i.e. the time corresponding to the pump-probe temporal overlap) is thus affected by some uncertainty, because each probe wavelength overlaps the pump pulse at a slightly differ-ent time (figure 2.3). Its polynomial estimation is therefore mandatory to correct data and locate a common zero for all kinetic traces.

(42)

Figure 2.3: A positive chirped probe causes a negative chirped signal. On the right, an example of the polynomial fit of the probe dispersion obtained by Glotaran.

(43)

2.3 Transient IR Absorption Spectroscopy

2.3 Transient IR Absorption

Information on the electronic excited states ladder is easily accessible by means of transient absorption spectroscopy in the Visible range; structural information is, instead, hardly achievable. Broad transient Vis bands span over hundred wavelengths in the ten thousand wavenumber region, they often overlap and are usually featureless and not informa-tive about the acinforma-tive vibrational modes in excited states. The detection of structural changes in excited states is in-teresting for the characterization of meta-stable transition states along a reaction coordinate, as well as for the iden-tification of predominant vibrational modes in non-radiative deactivation pathways. Non-radiative transitions always pro-ceed through vibrations (e.g. internal conversion through vibronic coupling or thermal relaxation through coupled vi-brations with the bath) and often this leads to a departing from the Franck Condon region. Structural information can be retrieved from Infrared and Raman Spectroscopies, which are sensitive to different microscopic properties: in the in-frared any change in the orientation, intensity and energy of the transition dipole moment associated to a specific nu-clear rearrangement is probed. Complementary to infrared absorption, vibrational modes can be probed also by Raman effect, which, similarly to an electronic spectroscopy, involves visible beams and is sensitive to any change in the electronic

(44)

2.3 Transient IR Absorption Spectroscopy density distribution over the entire molecule, namely to any change of the molecular polarizability anisotropy. While the transition dipole moment is a vector, the polarizability is a tri-dimensional tensor which can be considered as a rough estimate of the molecular shape.

2.3.1 Set-up

Ti:Sapphire Laser System λ=800 nm; P = 3W; Δt =35 fs; 1 KHz VIS Delay Line Computer NOPA (VIS pump) OPA (IR probe) TOPAS (IR pump) IR Delay Spectrom. MCT

Figure 2.4: Schematic view of the set-up for Transient IR Ab-sorption, 2D-IR, Transient 2D-IR and Labeling measurements.

The experimental set-up used for time resolved infrared measurements has been extensively described before.68,69

Briefly, a portion of the output of a Ti:Sapphire oscilla-tor/regenerative amplifier, operating at 1 kHz, centered at

(45)

2.3 Transient IR Absorption Spectroscopy

800 nm, with pulse duration less than 50 fs (Legend Elite, Coherent), was split in order to generate the mid-IR probe and the Visible (VIS) pump. The infrared probe had a spec-tral width of 200 cm−1 _{in the 6 µm region and was obtained}

from difference frequency generation in a AgGaS2 crystal of

the signal and idler beams from a homebuilt two-stage opti-cal parametric amplifier (OPA). The mid-infrared output was further split into two beams of equal intensity, which were respectively used as probe and reference and focused in two different spots on the sample. The VIS pump at 400 nm was generated by frequency doubling another portion of the laser fundamental beam in a BBO crystal, while the excitation at 540 nm was generated by a home-made non-collinear opti-cal parametric amplifier (NOPA). At both excitation wave-lengths the VIS-pump pulse was attenuated to 100-300 nJ at the sample and focused in a 150 micrometers diameter spot. The polarization of the pump was set to the magic an-gle with respect to that of the probe by rotating a λ/2 plate. A moveable delay line along the pump path made it possible to vary the time delay with respect to the probe up to 1.8 ns. After the sample, both probe and reference were spectrally dispersed in a spectrometer (TRIAX 180, HORIBA Jobin Yvon) and imaged separately on a 32 channels double array HgCdTe detector (InfraRed Associated Inc., Florida USA) under nitrogen. All spectra were recorded in two spectral windows covering from 1450 to 1800 cm−1 _{with a resolution}

(46)

2.3 Transient IR Absorption Spectroscopy of 6 cm−1_.

2.3.2 Data Analysis

S₀ S1 pum p 1 2 1500 1550 1600 1650 1700 1750 1800 -1 0 1 2 3 Δ A ( m O D ) wavenumbers (cm-1₎ 1 2 2 1500 1550 1600 1650 1700 1750 1800 -1 0 1 2 3 Δ A ( m O D ) wavenumbers (cm-1₎ 1 1 2 2

Figure 2.5: Transient IR absorption signals: negative signals (1) correspond to the bleaching caused by the pump of the vibrational modes in the ground state; pump absorption is followed by vibra-tional cooling on the electronic excited state in few picoseconds (undulated arrow); positive signals (2) rise from excited state IR absorption (ESA).

As reported for transient Vis measurements, all data were first accumulated, downloaded and saved through the LAB-VIEW programmed interface and then averaged over mul-tiple scans in Matlab. Once the two-way matrix was

(47)

ob-2.3 Transient IR Absorption Spectroscopy

tained, the spectral and kinetic analysis was done using both Glotaran and Origin softwares. The direct visualization of the two-dimensional data obtained from Glotaran highlights the striking different probe dispersion in the infrared, where it is almost flat, with respect to the visible region.

In average, the transition dipole moment of vibrational modes is about one order of magnitude smaller than that of electronic transitions. Then highly concentrated samples need to be prepared for infrared spectroscopy (∼ 10−2_M_).

The sample cell consists of two 1 mm thick fluorite windows, separated by a Teflon spacer of 50 or 100 µm. The small optical path length avoids self-absorption phenomena of the Vis pump and limits the pump beam attenuation along with the optical path. Therefore, the best signal-to-noise ratio is achieved by keeping concentration close to the solubility limit but avoiding in the meanwhile scattering over-saturated solution.

The frequency resolution does not exceed 6 cm−1 _{due to}

the limited number of active elements (32) in each MCT ar-ray. The spectral calibration is obtained by measuring the absorption spectrum of a standard polystyrene sample for all the frequency regions considered. Calibration is conducted taking as a reference its 0.2 cm−1 _{resolved FT-IR spectrum.}

(48)

2.4 Two-dimensional IR Spectroscopy

2.4 2D-IR Spectroscopy

Two-dimensional infrared (2D-IR) spectroscopy is a ground state technique that provides a powerful tool to study tran-sient molecular structure and dynamics. As a vibrational spectroscopy, it directly watches at vibrating molecular bonds and how those vibrations interact with one another and with the sorroundings.

Similarly to the two-dimensional techniques first developed and widely applied in nuclear magnetic resonance (NMR), 2D-IR spectroscopy spreads a vibrational spectrum over two frequency axes. From such a 2D spectrum one learns how ex-citation of a vibration influences all other vibrations (within a given spectral window); if short infrared pulses are used, direct information on the system time evolution becomes ac-cessible.

Spectral features, namely frequencies, amplitudes, line-shapes, and their evolution in time are used to understand structural connectivity in space and time, thus offering new possibil-ities for studying molecular structure and dynamics. If we do precede a visible pulse, in resonance with some molecular electronic transition, to the 2D-IR pulse sequence, we realize a non-equilibrium variant of the technique, from which we learn about the molecular dynamics in the excited state.

(49)

In the following, we briefly outline two different approaches to the realization of a 2D-IR experiment:

• the frequency domain: this it is the method (also known as "dynamic hole burning") that was realized first.70_It

makes use of a conventional pump-probe geometry with a narrow band pump selected by a Fabry-Perot interfer-ometer that scans the frequency across the broad band of the IR pulse. Two electric fields from the pump pulse impinge simultaneously and collinearly on the sample; a second delayed pulse, the probe, induces a collinear response signal. The probe field acts as carrier wave of the weak signal (self-heterodyne detection), which is spectrally dispersed and recorded in the frequency domain. The two axes of the absorptive 2D spectrum correspond to the frequencies of the pump and probe pulses, respectively. This is the experimental approach adopted in this thesis work; the used set-up will be described in details in the next section.

• the time domain: it is a Fourier Transform method also known as "Photon-Echo". Three beams impinge on the sample from different directions with district ar-rival times: the first two pulses act as the IR pump, the 3rd_{one is the probe. The response signal is emitted in a}

direction that does not coincide with any of the incom-ing beams; the heterodyne detection can be realized by

(50)

2.4 Two-dimensional IR Spectroscopy adding an external local field coincident with the signal. A two-dimensional interferogram is collected by scan-ning the time delay between the first two pulses and that between the third and the local field . A double Fourier transformation gives back the pump and probe frequencies.

Advantages and disadvantages of the methods and the tech-nical details are deeply discussed in ref.71_.

2.4.1 Set-up

2D-IR spectra were recorded on the system of figure 2.4 in the frequency domain. The pump and probe mid-IR beams were generated by two optical parametric amplifiers, as previ-ously described (see paragraph 2.3.1). From both parametric amplifiers a gaussian beam of 100 fs, 200 cm−1_{broad and}

cen-tered at 6 µm was obtained. The spectral bandwidth of the intense IR pump was narrowed to 15-17 cm−1 _{by means of}

a Fabry-Perot interferometer (etalon); correspondently, the pulse duration increases to about 800 fs (τ) and the pulse energy after the interferometer was about five times smaller than before. The pump beam was finally sent to a moveable delay line; its polarization was set to the magic angle with respect to the probe by means of a λ/2 waveplate. The fo-cusing at the sample and the detection system was the same

(51)

as that reported in section 2.3.1 for the transient infrared absorption set-up.

2.4.2 Data Analysis

1610 1620 1630 1640 1650 1660 1610 1620 1630 1640 1650 1660 probe frequencies (cm-1₎ p u m p f re q u e n ci e s (c m -1) 1+2 5 |0,0> |1,0> |1,1> |0,1> |0,2> |2,0>

S

₀ 2 1 3 4 8 6 5 7 5+6 3 7 8 4 1

Figure 2.6: 2D-IR spectrum in presence of strongly coupled modes (Phenol Blue in dichloro-methane): negative signals are in blue and correspond to bleaching and stimulated emission pro-cesses, positive signals are in red and correspond to excited state absorption.

The two axes of a 2D-IR spectrum correspond to the pump and the probe frequencies. Each 2D spectrum is measured at a given pump-probe delay and results from the interpolation of side-by-side 1D spectra of the probe acquired for different

(52)

2.4 Two-dimensional IR Spectroscopy frequencies of the band pump. The spectral narrow-ing of the pump causes a Fourier intrinsic loss in the time resolution.

Two types of measurements on this set-up are possible. If both pump and probe beams are broad-band, i.e. if the Fabry Perot is removed from the pump pathway, experiments can reach the best time resolution, but it will be impossible to selectively excite the individual modes. On the other side, if a narrow-band pump is used, dynamics within the first picosecond will be lost, but it will be possible to follow the time evolution of the individual modes and to disentangle the inter-mode couplings.

(53)

2.5.1 Transient 2D-IR Spectroscopy

2.5 5

th

order spectroscopies

In the case that a typical 3rd order non linear spectroscopic technique, like 2D-IR, is used to probe the properties and the evolution of an electronic excited state prepared by a visible pulse, the overall experiment can be described as an example of 5th order spectroscopy. In such an experiment, two electro-magnetic fields are contributed by the visible pump pulse; two more fields are involved in the action of the second pump, and the fifth is brought by the probe pulse.

In the following, we consider two different 5th_order

spec-troscopic techniques that use, with a different time ordering, one visible pump, one infrared pump and one infrared probe. The set-up reported in figure 2.4 was properly adapted in order to accommodate the pulse sequence for the appropri-ate three beams experiment. In the last part of the chapter, a third 5th order experiment, the Time Resolved Stimulated Raman, will be described. This set-up was built in collabora-tion with Dr. Tomasz Kardas from the University of Warsaw.

(54)

2.5.1 Transient 2D-IR Absorption: Vis

pump-IR pump-IR probe

As previously discussed, electronic excited state dynam-ics can be studied by visible and infrared transient spectro-scopies. It is also apparent that adding a second dimension to transient spectra opens the access to relevant informa-tion, concerning the presence of hidden peaks as well as of cross peaks that may be important to unveil the nature of the excited states. In transient 1D infrared spectra, ground state bleachings are often hidden by positive excited state bands. In the electronic excited state some vibrational modes can show very large frequency shift with respect to their fre-quency position in the ground state, as a result of large vi-bronic coupling (i.e. the coupling of the dipole moment as-sociated to the vibration with the electronic dipole moment of the electronic excited state). The analysis of active vibra-tional modes in the electronic excited state allows monitoring structural changes associated with the relaxation. Breden-beck et al.72,73 _{were the first to add a second dimension to}

transient IR spectra. They applied a 2D-IR pulse sequence after a visible excitation pulse, so that, repeating the exper-iment at different delays between the visible pump and the infrared sequence, they obtained 2D spectra of an evolving non-equilibrated system.

(55)

2.5.1 Transient 2D-IR Spectroscopy IR probe IR pump _time VIS pump t _IR t _VIS

Figure 2.7: Pulse sequence in a Transient 2D Infrared experi-ment.

In our experiment, based on the Transient Infrared and 2D-IR set-up (figure 2.4), we combine the visible pump gen-erated by the home-made NOPA, the infrared pump from the commercial TOPAS, followed by the Fabry Perot inter-ferometer, and the infrared probe from the home-made OPA. Measurements were conducted at 400 and 520-540 nm exci-tation wavelengths. The exciexci-tation pulses at 400 nm were obtained from second harmonic generation (SHG) of a por-tion of the laser fundamental in a BBO crystal, resulting in a bandwidth of 6 nm. The excitation wavelength around 520-540 nm was finely tuned by controlling the NOPA gen-eration, so that it was centered at half height of the steep slope on the red side of the carotene static Vis absorption spectrum; its bandwidth was 20 nm. A λ/2 plate was used to set the polarization of the visible beam at the magic angle with respect to IR beams, whose relative polarization was set parallel. the pulse energy of the visible pump was attenuated

(56)

2.5.1 Transient 2D-IR Spectroscopy to provide 100-300 nJ at the sample, in a focused spot of 150 µm diameter.

As shown in figure 2.7, two time delays can be varied: we fixed the IR delay at the maximum of the pump-probe IR signal (tIR= 500 fs) and scanned the delay tV IS of the

visi-ble pump with respect to both IR beams. In this way, single snapshot (in the form of 2D infrared spectra) are caught dur-ing the relaxation of the electronic excited state.

The necessary steps involved in measuring transient 2D spectra can be summarized as follows:

1. optimization of the narrow-band IR signal at the ground state (VIS pump blocked). The IR pump is chopped to extract the difference absorption spectra. Fix the IR delay between pump and probe at the maximum of signal intensity.

2. optimization of the transient IR signal (IR pump blocked). The VIS pump is chopped and the linear transient in-frared difference spectrum is measured. At this step, excited state vibrations are visible: take note of their frequency position.

3. repositioning of the chopper in the IR pump path. The central frequency of the excited state vibration will be the target position of the narrow-band IR pump. Close

(57)

the sample chamber and flux nitrogen for at least 15 minutes. Open both VIS and IR pump beams and look for the difference signal at the excited state vibrational frequency.

Generally, expected signals are weak, obviously depending on concentration and on the spacer used in the cell (i.e. the optical path length). A great stability of the baseline, which depends on the laser source stability and on the quality of the IR probe generation, is strongly recommended to increase the signal to noise ratio together with the lack of humidity inside the sample chamber.

Excited state 2DIR spectra result from a double difference absorption signal: ∆∆A = ∆AV ISon

IR − ∆A

V ISof f

IR . Using

a single chopper placed along the IR pump path, the dou-ble difference spectra is post-processed by subtracting the background, in other words by subtracting the 2D spectrum collected at negative delays (tV IS = −10ps) from the ones

measured at positive delays. Note that at negative delays both IR beams impinge on the sample before the Vis pulse. Hence, in T2D-IR spectra, the ground state signals result with inverted signs: the bleaching and the stimulated emis-sion are positive, while vibrational excited state absorptions are negative.

(58)

2.5.1 Transient 2D-IR Spectroscopy S₁ |0,0>’ |1,0>’ |0,1>’ |0,2>’ |2,0>’ |0,0> |1,0> |1,1> |0,1> |0,2> |2,0> S0 |0,1>’ |1,0>’ |1,0> |0,1> probe frequencies (cm-1₎ pu mp fre qu en ci es (cm -1)

Figure 2.8: Exemple of transient 2D-IR spectrum: coupled

modes in the ground state loose their coupling in the excited state and shift to opposite directions.

Usually diagonal peaks are easily identified by compari-son to the FT-IR and transient 1D-IR spectra, however, the interpretation is not so obvious in the off-diagonal regions, especially in highly coupled systems. Since three beams are used, one could decide to fix the first delay between the Vis and the IR pump and scan the IR pump-probe delay, as re-ported by Bredenbeck et al.72 _{If the electronic excited state}

lives longer than the cooling on its vibrational manifold, a complete spectral diffusion analysis of an excited state vibra-tional mode is possible and the solvent rearrangement around the excited solute molecule can be monitored.

(59)

2.5.2 Excited State Vibrational Labeling

2.5.2 Excited State Vibrational Labeling: IR

pump-Vis pump-IR probe

IR probe

VIS pump _time

IR pump

t ₂

t ₁

Figure 2.9: Pulse sequence in a Vibrational Labeling experiment.

In this technique, the order of the Vis and IR excitations is inverted with respect to the T2D-IR pulse sequence (fig-ure 2.9). A first narrow IR pulse excites, or in other words "labels", a vibrational mode in the ground state. Right after it (t1), a VIS pump pulse transfers the system from a

non-thermally equilibrated ground state to an electronic excited state. In this way, the vibrational population distribution of the ground state is directly projected into the electronic excited state (figure 2.10). Finally, the IR broad band probe pulse interrogates the excited system. In order to maintain memory of the labelled vibration, the time-of-arrival of the probe with respect to the IR pump (tIR=t1+t2) has to be set

within short lifetime of the vibration in the excited or in the ground state. In this case, the probe unambiguously identi-fies the frequency of the vibrational mode in the electronic

(60)

2.5.2 Excited State Vibrational Labeling excited state. S₀ S1 IR VIS probe probe probe frequencies (cm-1₎ pu mp fre qu en ci es (cm -1) S₀ S₁

Figure 2.10: A visible excitation projects the out-of-equilibrium population of the ground state to an electronic excited state within the vibrational lifetime.

The excited state signal is expected in the off-diagonal re-gion, as shown in the example sketched in figure 2.10. The IR pump is tuned at the frequency of the ground state vibration, while the broad IR probe reveals simultaneously the ground state vibration (on the diagonal) and the shifted electronic excited state vibration (at higher frequencies in this case). The intensity of the excited state vibration in the off-diagonal region is enhanced by the vibronic coupling and decays with increasing t1+t2 delay (see figure 2.9), according to the

life-time of the vibrational mode. The labeling spectra reported in this thesis were measured at fixed delay tIR= t1+t2= 1

(61)

2.5.2 Excited State Vibrational Labeling

ps, scanning the Vis pump-IR probe t2delay within the tIR=

1 ps delay. Background spectra were acquired without the Vis pump and correspond to equilibrium ground state 2D-IR spectra recorded at 1 ps. We observe an increase of the off-diagonal peak intensity when the Vis pump was moved closer in time to the IR pump. The relative polarizations of the three beams were set parallel. By keeping the Vis pump IR pump delay fixed at 500 fs we avoided artefacts due to the temporal overlap to the pump, still maintaining a reasonably good signal intensity. Indeed, for small values of t1, the

out-of-equilibrium population in the ground state is frozen by the Vis pump before vibrational cooling takes place.

This technique was successfully applied for the first time by Bredenbeck et al.72,74 _{to assign the symmetric and}

asym-metric stretching modes of equatorial and axial C=Os of a bipyridyl- Rhenium(I) complex upon metal to ligand charge transfer (MLCT) transition. Here, we re-propose tho method to unambiguously assign the C=O stretching vibration of ca-rotenoids in the electronic excited states and, consequently, to confirm the assignment of the largely upshifted chain C=C stretching.

(62)

2.5.3 Transient Stimulated Raman Spectroscopy

2.5.3 Transient Stimulated Raman

Spectroscopy

In the literature this technique is better known as Fem-tosecond Stimulated Raman Spectroscopy (FSRS). First de-veloped by R. A. Mathies at the University of California, Berkeley, in 200375,76_{, it consists of a Stimulated Raman}

ex-periment in the electronic excited states. The word "Fem-tosecond" refers to the time resolution that can be reached when the two simultaneous Raman pulses that impinge on the sample are delayed with respect to the visible actinic pulse. Time-resolved Stimulated Raman spectra are thus recorded at different "actinic pump - stimulated Raman probe" delays. Since its first implementation, this technique has been applied for a variety of topics: molecules involved in photosynthesis, like β-carotene77_{, molecules involved in the}

mechanism of vision78,79_{(rhodopsin), photo-induced spin}

cross-over in iron complexes80 _{and, recently, excited state proton}

transfer81_{. In 2013, thanks to Dr. Kardas, we built a FSRS}

set-up in our laboratory, thus implementing the transient lin-ear and two-dimensional infrared spectroscopic instrumenta-tion with the complementary transient Raman counterpart.

The Raman scattering effect can be seen in a classical approach as an inelastic collision between a photon and a molecule, in which a small part of the overall collision energy

(63)

2.5.3 Transient Stimulated Raman Spectroscopy

can be left on the molecule or given to the photon. Micro-scopically, a molecule is defined as "Raman active" when the applied external electric field causes a variation in its polariz-ability at least along one of its molecular coordinates. From a macroscopic point of view, instead, the overall polarization

~

P reacts instantaneously to the light perturbation accord-ingly to the electric susceptivity of the solution. Although spontaneous Raman effect is usually described as a linear spectroscopic technique, since the recorded signal intensity is ultimately proportional to the intensity of the applied ex-citation beam, all Raman techniques are better described as 3rd_{order non-linear spectroscopies.}82,83 _{In the following, we}

will focus on the differences between the spontaneous and the stimulated Raman spectroscopies to then up-grade to the ex-cited state time-resolved technique.

• Spontaneous Raman Scattering

Only one visible non-resonant beam is focused on the sample. The light is then incoherently scattered in the entire solid angle and detected in a back-scattering ge-ometry, orthogonally or possibly with an integrating sphere. On both sides of the strong elastic scattering Rayleigh line, the weak Raman lines appear, separated from the Rayleigh line by vibrational energy quanta. In figure 2.11, the Raman intensity is plotted versus the vibrational difference frequencies ∆ω = ωscatt− ω0,

(64)

2.5.3 Transient Stimulated Raman Spectroscopy expressed in wavenumbers. On the Stokes side photons are scattered at lower frequencies than the Rayleigh line: ωscatt < ω0 because part of the photon energy

is left to the molecule. On the anti-Stokes side, in-stead, an excess of energy initially in the molecule is transferred to the photon which is scattered at higher frequencies: ωscatt > ω0. Exchanged quanta between

molecules and photons correspond to the characteris-tic vibrational energies of the investigated molecular system, therefore, Stokes and Anti-Stokes transitions are specular with respect to the Rayleigh line. The two branches differ in intensity because of the Boltz-man factor: while Stokes transitions depart from the ground state, i.e from the minimum energy configu-ration where almost all the molecules are at thermal equilibrium, the Anti-Stokes intensities depend on the population of hot vibrational levels (at room tempera-ture only modes with E < 200 cm−1 _{are active). Note}

that from the perturbation theory, virtual states |n1>

and |n2>, which are not stationary eigenstates of the

system, can instantaneously exist as superposition of all the eigenstates.

(65)

2.5.3 Transient Stimulated Raman Spectroscopy Stokes Anti-Stokes -200 -100 0 100 Δω (cm-1₎ 200 Rayleigh Ground State

Stokes Rayleigh Anti-Stokes

|v0> |v1> |n1> |n2>

Figure 2.11: Schematic view of a spontaneous vibrational Ra-man spectrum and corresponding level diagrams with Stokes, anti-Stokes and Rayleigh transitions.

• Stimulated Raman Scattering

As reported in figure 2.12, two beams from different di-rections, the Raman pump (red) and the probe (blue), impinge on the sample at the same instant generating a vibrational Raman coherence in the electronic ground state of the molecule. Due to the phase-matching con-dition (~ksignal = ~kpump− ~kpump+ ~kprobe), the signal

Excited state dynamics of carbonyl carotenoids investigated by ultrafast vibrational and electronic spectroscopies.

Contents

Introduction

1 The Photosynthesis

1.1 The Light-Harvesting Proteins

1.1.1 The Peridinin-Chlorophyll a Protein

1.1.2 Energy-Transfer mechanisms in

Light-Harvesting Proteins

1.2 Electronic properties of

Carotenoids

1.2.1 Carbonyl Carotenoids

2 Spectroscopic

techniques and set-ups

2.1 Steady-state UV-Vis and IR

Absorption

2.2 Transient UV-Vis Absorption

2.2.1 Set-up

2.2.2 Data Analysis

2.3 Transient IR Absorption

2.3.1 Set-up

2.3.2 Data Analysis

2.4 2D-IR Spectroscopy

2.4.1 Set-up

2.4.2 Data Analysis

S

2.5 5

order spectroscopies

2.5.1 Transient 2D-IR Absorption: Vis

pump-IR pump-IR probe

2.5.2 Excited State Vibrational Labeling: IR

pump-Vis pump-IR probe

2.5.3 Transient Stimulated Raman

Spectroscopy